Re: debian-installer (was Re: Unofficial Sarge Release Issues (Updated for August))
On Wed, 6 Aug 2003, Josip Rodin wrote:
> On Wed, Aug 06, 2003 at 10:22:22AM -0500, Drew Scott Daniels wrote:
> > > > - The debian-installer (d-i) is in what I would call almost a releasable
> > > > state, but that's just my opinion. See debian-boot for information.
> > > Without it being on the CDs, it's far from a releaseable state...
> >
> There's bootability only on i386 right now, AFAIR, and even that was
> worthless when I last tried it two weeks ago -- the CD booted fine, and
> had all the packages, but I couldn't tell the installer to actually use
> the file:/cdrom URL for packages, it insisted on network (regardless of
> which image I selected).
>
I missed reading bug #201537 [1]. A new thread [2] on debian-boot may
prove interesting.
> > If this is an issue, then it's worth noting that
> > http://lists.debian.org/debian-devel/2002/debian-devel-200207/msg01479.html
> > says: "If you don't have a functioning installer by release, then sure,
> > you'll release, but you won't be installable when you do." This means
> > that, like with Woody (afaict), Sarge may be released without a functional
> > installer for some of the "supported" architectures.
>
> Yes but I'm sure nobody considers a regression from ~ 11 working install
> procedures to ~ 0 :)
>
What can I say, other than it somewhat worked for me. It looks like d-i is
quite "close" on several architectures [3], but yes, close isn't
releasable.
How much of a regression would be permissible? I think down to even just
i386 would be grudgingly accepted if the other arches were given "enough"
time, but this is all speculation and conjecture as there seems to be some
time left before freeze, and lots of time left before release [4]. Now
that I think of it though, if Adrian Bunk had his way it may be acceptable
to have a release without an installer and shortly after have a new
release hopefully with an installer.
Looking through the TODO list [5] I don't see any RC bugs, but quite a few
issues that should probably be fixed before release. I'm a bit surprised
to see bug 201537 (submitted 19 days ago). Is that the only thing holding
back the installer from being in a releasable state (apart from sprinkling
of udeb RC bugs)? Even if it is by my standards, it may not be by the
standards of others.
I guess it would be appropriate to wait until after Sept 25th's d-i camp
to finish before expecting d-i to be "ready". That said, I think I'll talk
to debian-boot about things that might be considered at the d-i camp.
[1] http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=201537
[2] http://lists.debian.org/debian-boot/2003/debian-boot-200308/msg00059.html
[3] http://cvs.debian.org/*checkout*/debian-installer/doc/html/ports-status.html
shows i386, ia64 and powerpc usually work. hppa, m68k and s390 need more
work (or maybe just some testing?). Some of the other archs have patches
in the BTS for the bugs holding them back from being almost working.
[4] By my estimates anyway.
[5] http://cvs.debian.org/debian-installer/doc/TODO?rev=HEAD&content-type=text/vnd.viewcvs-markup
Drew Daniels
Reply to: